Overview & result
I used Jev to solve a problem every agent eventually runs into: reading logs. 22.8M lines, and running an LLM on every one would've cost $1,120. Tocsin groups them into 11,812 repeating patterns, then asks Jev about each pattern once. 6 minutes, 64 cents, 123 patterns that actually needed to be looked at. http://github.com/TPAteeq/tocsin The paging policy is just a prompt. You tell it what should wake someone up at 3 am and what's just another log line.
How Jev fits in the loop
- Numbers, IPs, UUIDs, hex values and emails are masked so that log lines collapse into patterns.
- Jev is asked about each pattern once, against a paging policy you write in plain English.
- Every line is routed to page, ticket or log according to its pattern's verdict.
- Verdicts are cached, so a repeated pattern costs nothing more.
How to reproduce
- Inspect the original showcase and source material at https://github.com/TPAteeq/tocsin
- Verify the bounded prompt and input candidate schema configured for Jev
- Benchmark decision latency and classification accuracy against baseline models
Why this build matters
It applies Sentry's group-then-alert idea to raw logs, and the author's write-up includes the benchmark run where it lost.
Limitations
- Lines longer than 16 KiB are truncated before templating.
- One process per cache file: two writers on the same file overwrite each other's entries.
- jev-latest is an alias, so pin a model version if verdicts must stay fixed; webhook alerts are dropped after four retries.